Shift, a division of the German startup Microagi, is using private chefs to collect first-person video data for training robots. During a recent visit, chef Ollie recorded his movements while preparing a meal, with the intention of improving robotic culinary skills. The data collected includes hand movements and task execution, which Shift believes is essential for developing humanoid robots capable of performing complex tasks at home.
The initiative is part of a broader trend in egocentric data collection, where individuals record their actions for AI training. Shift expanded its services in San Francisco by offering free home-cooked meals in exchange for recording the process. This approach mirrors a similar program in New York City, where the company provided free cleaning services in return for video data. The company's CEO, Bercan Kilic, emphasized the goal of creating a marketplace where contractors can monetize their data contributions while advancing AI models.
Kilic envisions a future where home robots, like vacuum cleaners in 2020, will be available at reasonable prices and capable of handling most household tasks. However, he acknowledges the challenges of ensuring safety and avoiding unintended consequences, such as a malfunctioning chef-bot endangering humans or pets. While he remains optimistic about the potential for robotics to improve living conditions, he also notes the persistent issues of wealth inequality and homelessness that may not be fully resolved by technological advancements.
Source: wired
